Hi!
Is there a good way of registering images by only considering the x and y translation?
The registration does not need to be perfect as i only want to use the x and y translation from the transformation matrix to simplify things later in the program.
In the current code, i use SURF to detect and match features which gives me a transformation matrix which also includes shearing, scaling and rotation. This however seems to have an effect on the values of x and y translation as when i use those to values alone to match the before and after image, they are a bit off.
I have provided the before and after images.
Any suggestions?
Regards,
Anders Holmberg